Tuesday, November 8, 2011


What is one to do, now that we’ve learned that early detection of cancer isn’t always as important as we’d thought? Should we get screened or not? Some screening tests, as for cervical and colorectal cancer, reliably lead to effective treatment, but serious questions about others, especially PSAs (for prostate cancer) and mammograms, are emerging.

I’m not writing here about the value of particular tests, though, but about testing per se. Healthcare experts are looking critically at testing itself these days, no doubt because healthcare’s gotten so ridiculously expensive. They’re questioning exactly what’s gained, analyzing cost-benefit ratios. An example is the CT lung-scanning of smokers to screen for cancer. Besides being costly, CT scans, like all tests, are subject to “false positive” results that can encourage unnecessary biopsies and woe. And by the way, they're also subject to false negatives.

Why can’t we have tests without false positives and negatives? We’re far beyond reading chicken entrails at this point, but we’re still not perfect. And you know what? Medical science itself will never be perfect, and not just because we lack some tool. Nobel Prize winners since Werner Heisenberg have reaffirmed that uncertainty isn’t some cosmic condiment, but is, in fact, the only item on the menu.

Recognizing that complete security is a myth, then, how much of it are you willing to buy? In healthcare, will you pay a hundred dollars for an x-ray that carries ninety percent certainty? Or would you prefer the ninety-nine percent certainty of a thousand-dollar CT scan? Or perhaps for two thousand dollars, you’d go for the ninety-nine-point-nine-percent certainty of an MRI. (Whatever your choice, of course, there are no guarantees.)

What’s my preference? My most comfortable approach would be the one that feels, well, most comfortable, after trying a few on. Why not explore your U.Q., your Uncertainty Quotient? People mattering in healthcare includes seriously, responsibly, consciously realizing less our wants and more our needs.

1 comment:

  1. And expressing it in percents is hard for people to grasp, or at least to manipulate in their heads - you get wildly different responses depending on how the Q is framed.

    Somewhere recently I saw an image of a grid, to use to communicate about prevalence, & false positives & false negatives in a diagnostic test. The grid works better than percents, since we're a lot better at grasping this stuff visually. This wasn't the page, but if you google
    "NATURAL FREQUENCIES GRID" diagnostic "false positive"
    you'll get some hits that describe it.